the science of the total environment, 17 (1981) 165--196

The Science of the Total Environment, 17 (1981) 165--196 Elsevier Scientific Publishing Company, Amsterdam -- Printed in The Netherlands

165

A S T A T I S T I C A L A S S E S S M E N T O F THE F O R M O F T R A C E M E T A L S IN O X I D I Z E D E S T U A R I N E S E D I M E N T S E M P L O Y I N G C H E M I C A L E X T R A C T A N T S

SAMUEL N. LUOMA*

U.S. Geological Survey, 345 Middlefield Road, Menlo Park, CA 94025 (U.S.A.)

G. W. BRYAN

The Laboratory, Citadel Hill, Plymouth, PL1 2PB (United Kingdom)

(Received February 27th, 1980; accepted in final form August 19th, 1980)

ABSTRACT

The thin layer of oxidized sediment at the sediment--water interface plays an important role in the chemical and biological interactions of trace metals in estuaries. Chemical extractants can be useful in defining trace metal interactions in sediments. Extractions may aid in determining the abundance of the operationally-defined forms of substrates which are the most active in binding metals in sediments; in operationally-defining an "extractable" phase of trace metals in sediments; and in providing information necessary for determining the bioavailability of sediment-bound metals. Comparison of 10 techniques for metal and substrate extractions among the sediments of 19 estuaries from south and west England indicate substrate characterization is best accomplished by Fe and Mn extractions with acid ammonium oxalate or 1NHCI, and humic substances extraction with 0.1N NaOH or 1N ammonia. Partial fractionation of trace metals is best accomplished with 1N HC1. Statistical relationships indicate the extractable phase of Fe is more important than total Fe in binding Ag, Cd, Cu, Pb and Zn in oxidized sediments, and the operationally-defined humic substance fraction of organic materials is highly important in binding Ag and Cu. Statistical analy.sis within specific subsets of data indicate trace metals are partitioned among several substrates in most sediments, the substrates compete with one another for the metals, and the outcome of the competition is strongly influenced by the concentrations of the different substrates in the sediment.

INTRODUCTION

S e d i m e n t s c o n s t i t u t e one o f t he m o s t c o n c e n t r a t e d s inks o f t race me ta l s in aqua t i c e n v i r o n m e n t s . Wi th in sed imen t s , me ta l s m a y p a r t i t i o n a m o n g several d i f f e r e n t t y p e s o f b i n d i n g si tes ( subs t ra tes ) . This p a r t i t i o n i n g is g rea t ly i n f l u e n c e d b y the p h y s i c a l a n d c h e m i c a l cha rac te r i s t i c s o f t he sedi- m e n t . P a r t i t i o n i n g a m o n g d i f f e r e n t p h y s i c o c h e m i c a l f o r m s p l ays an i m p o r t a n t

*Member of the Society for Environmental Geochemistry and Health.

0048-9697/81/0000--0000/$02.50 ©1981 Elsevier Scientific Publishing Company

166

role in determining the biological significance of sediment-bound metals and in determining metal exchange between sediments and water (Luoma and Jenne 1977; Luoma and Bryan, 1978; 1979).

Determining metal partitioning in the layer of sediment at the interface of the sediment bed and the water column is especially important in estuaries. The sediment--water interface plays several roles in the chemistry of trace metals: ( 1 ) T h e surficial layer of sediment is usually predominantly oxidized, and thus acts as a diffusion barrier for solutes migrating upward from reducing zones of sediment. Precipitation of manganese oxides and ferric hydroxides (with an associated coprecipitat ion of available trace metals) at the interface will control the exchange of trace elements between sediments and the water column (Hem, 1977; 1978). (2)Surface sediments on the bed of many estuaries exchange readily with suspended sediments (Conomos and Peterson, 1977); and (3 )The sediments at the water--sediment interface are more important to biological fauna than are subsurface sediments. Meiofauna live exclusively above the reduced zone in sediments. Suspension feeding and epibenthic organisms are exposed directly to metals in the environment of the sediment--water interface; and most macroscopic infauna use tubes, burrows or siphons to obtain the bulk of their food and oxygen (and thus their metal burden) from the sediment surface.

The trace metal chemistry of oxidized sediments such as those at the sediment surface is more complex than that of anoxic sediments. Sulfides dominate the processes controlling metal form in reducing sediments (Morel et al., 1975). However, in oxidized sediments, organic materials, carbonates and hydrous oxides of both Fe and Mn may all compete for the binding of trace metals. (For the purposes of this discussion we will assume the term binding encompasses adsorption, complexation, coprecipitation and ion exchange.) Clays and other silicate mineral surfaces also have some capacity for binding metals. However, the strength of metal association with clay surfaces is weak relative to metal associations with substrates which would complete for binding in oxidized sediments. Thus, the most likely role of the clays in such sediments is that of a carrier for the substrates which bind metals more strongly (Jenne, 1977).

The binding substrates themselves occur in a variety of forms; and the forms of the substrates in any given sediment greatly influence the strength of metal--substrate binding processes. In estuarine sediments, major classes of organic materials include humic substances, bacteria, bacterial metabolites and refractory, non-viable organic material (e.g. cellulose and lignins). Oxides of Fe exist in layers on particle surfaces or as particles of mineralogic Fe. The crystallinity of the Fe oxides varies over a cont inuum from highly amorphous to highly crystalline. Manganese may occur as Mn oxides of varying structure or Mn carbonate. In turn, both Fe and Mn may also occur associated with each other or with other binding sites.

Mathematical models may be the most realistic approach to predicting metal partitioning in sediments of varying character (analogous to models of metal speciation in solution). Partitioning models would require constants

167

describing the strength of metal--substrate binding, and values for the relative abundances of the different forms of the different substrates present in natural sediments. Extraction of sediments with established geochemical techniques may provide information useful in such modeling efforts. Extrac- tions could be used to describe the abundances of operationally-defined forms of substrates in different sediments. Statistical relationships between trace metal concentrations and the different substrate forms defined by the extractants may also indicate which of these substrate forms to employ in models (or the development of constants for the models).

Most recent studies employing chemical extractions have attempted to directly define metal partitioning, often implying a degree of specificity for extracting metals from single substrates which probably cannot be obtained with most procedures. Studies with chemically well-defined sediments suggest extractants which selectively remove metals from a single substrate (much less from a given form of a substrate) are the exception rather than the rule (Luoma and Jenne, 1976; Guy et al., 1978). The useful information provided by the direct extraction of trace metals may be limited to determination of a general, operationally-defined "extractable phase" of the metals, and to providing information about the biological availability of bound metals (Luoma and Bryan, 1978; 1979).

In this paper, we compare the extraction of metals and substrates from surficial estuarine sediments by 9 commonly employed geochemical techniques, and determine statistical relationships between extractable metals and extractable substrates. Our objectives are: (1) To compare the usefulness of different extractants in determining an operationally-defined "extract~ able" fraction of metals in oxidized sediments. (2) To assess the characterization of the different forms of some substrates (Fe oxides, Mn oxides, the humate fraction of organic materials) by different extractants, both in terms of the relative concentration of substrate extracted and the statistical correlation of extractable fractions of substrates with trace metals; and (3) To test the hypothesis, implicit in previously reported statistical models coupling metal partitioning and metal availability to deposit feeders (Luoma and Bryan, 1978; 1979), that the partitioning of metals in sediments may vary with variations in the concentrations of substrates (i.e. that substrates actively compete for metals in a sediment and that competition is concentration-dependent).

METHODS AND MATERIALS

Sample collection and treatment Sediment samples were collected from 50 stations in 19 estuaries in south

and west England (for precise geographical locations see Luoma and Bryan, 1978). Estuaries and stations were chosen to give the widest possible vari- ability in physicochemical conditions to facilitate statistical analyses. Metal concentrations differed among stations by 1--3 orders of magnitude (Table 1);

1 6 8

,..1

r.r.l

0

Z

r~ I

~ •

z r~

o

E~

0

E~

~Q

o o o o ~ o 666

~ ~ 0 ~

0

e* Q;

0

169

substrate concentrations (Fe, Mn, humic substances, total organic carbon, carbonates) varied by 1--3 orders of magnitude; metal inputs included agri- cultural, urban, industrial and mining wastes; and the nature of the estuaries varied from broad, sandy sediment estuaries dominated by marine processes (e.g. Gannel River) to more protected estuaries with silt--clay sediments (e.g., Tamar River).

Samples of surficial sediment were scraped from the surface layers (~ 5 mm deep) of intertidal sediments at low tide. For comparison, subsurface sediments were collected at 5 cm depth below the redox interface at 7 of the stations. Within 4h or collection, sediments were sieved through 100/zm polyethylene mesh using diluted seawater, to remove particle size biases caused by large sand grains (de Groot, 1976). Following overnight settle- ment (within 24 h of collection) the 100 pm sediments were either extracted wet or dried at room temperature.

In the interest of developing sediment characterization schemes that were of maximum use in correlations with biological variables (Luoma and Bryan, 1978; 1979) and that allowed collection of statistically useful quantities of data, every effort was made to keep the chemical treatments as simple as possible. Subsamples of wet sediment were collected from a slurry using a pipette sampler while swirling. Each subsample was treated with a different extractant, i.e., samples were not extracted sequentially. Duplicate subsamples were washed with distilled water then dried at 80°C to obtain salt- free weights. Dry weights (250--550mg) seldom varied by more than 10mg between replicates. Salt-free weights were calculated for the dried samples from the salinity of the water used in the sieving procedure. Extractions were carried out in 20-ml glass scintillation-counting vials, which were shaken at frequent intervals. The extract was separated from the sediment by filtration under pressure through a 0.45-#m membrane filter. The filtrates were analyzed for Ag, Cd, Co, Cu, Fe, Mn, Pb and Zn by atomic absorption spectrophotometry. Background correction was used in Ag, Cd, Co and Pb analyses. Because of low concentrations in sediments, Cd is reported only in HC1 (HCl-soluble Cd equalled totals) while Ag is reported only in total~ HC1, ammonia and NaOH extracts.

Ex tractions Concentrated nitric acid (Total). Samples were wet ashed using the pro-

cedure of Bryan and Uysal (1978). Oxidation with concentrated nitric acid of sediments from the Tamar estuary resulted in nearly 100 percent extraction of Cu, Pb and Zn, and over 90 percent extraction of Mn and Fe compared to total sample dissolution in hydrofluoric acid (Bryan and Hummer- stone, 1971).

1N hydrochloric acid (HCI). Subsamples of 1 g air-dried sediments were extracted for 2h with 10ml of 1N HC1. Preliminary experiments showed similar concentrations of metal were extracted from air-dried and wet sediment. Weaker solutions of HC1 (0.1N) were neutralized by carbonates in the sediments, which greatly inhibited the action of the extractant.

170

25 percent acetic acid. Subsamples (2 ml) of wet sediment were extracted for 2 h in 20 ml of acetic acid. A slight neutralization of the acid (from pH 2.2 to pH 2.6 in the most extreme cases) was observed in sediments with the highest concentrations of carbonates. Dried subsamples of these sediments were re,extracted and titrated back to pH 2.2 with concentrated nitric acid shortly after initiation of the extraction. Only the extraction of Cu was detectably affected by the slight neutralization of the acetic acid. The copper values from the re-extraction were used in our data analysis.

0.4N ammonium oxalate in 0.4N oxalic acid at pH 3.3 (acid ammonium oxalate -- Schwertmann, 1964). Subsamples (2ml) of wet sediment were extracted for 2 h in 20 ml of extract and preserved for analysis by injecting 25pl of concentrated HNO3 into the filtrate.

0.1N hydroxylamine hydrochloride in O.01N nitric acid at pH 2 (hydroxylamine -- Chao, 1972). Subsamples (2ml) were extracted for 30min in 20 ml of extractant. Hydroxylamine was extremely sensitive to the presence of any CaCO3 in sediments (due to the poor buffering capacity of 0.01N nitric acid). Neutralization dramatically affected extraction of at least Cu, Zn, and Fe (Luoma, unpublished data); thus our hydroxylamine data was meaningful only at stations where carbonates were present in very low concentrations.

1N ammonium acetate at pH 7. Subsamples (2ml) of wet sediment were extracted for 2h. The filtrate was preserved for analysis with 25 pl concentrated HNO3.

0.1N sodium pyrophosphate. Subsamples (2 ml) were extracted for 2 h. 1N ammonia and 0.1N sodium hydroxide (NaOH). Subsamples (4ml)

were extracted for 1 week in 20 ml (~15:1 extractant/dry solid) of extract~ ant. The time of extraction was determined from preliminary experiments which indicated very slow metal and organic matter removal by both methods. The normalities of the two extractants were chosen to maximize the extraction of Cu and organic materials (Fig. 1).

The concentration of organic materials extracted (which we will operationally define as humic substances) was measured by light absorbance at 460 nm in a l~cm cell. This wavelength was within a broad band of absorbance observed in the extracts. In sediments from the Looe estuary 1 unit of absorbance/g sediment equalled 0.0005 g humic acid/g sediment (determined by repeated precipitation of the organics at pH ~ 2 and weighing the resi- due). Addition of FeC12 up to twice the highest Fe concentration found in any extract did not affect absorbance, suggesting all the absorbance was due to organic materials.

Carbon determination. Total organic carbon (TOC) was determined on a carbon analyzer by the difference between total carbon and inorganic carbon. Inorganic carbon was dissolved in boiling phosphoric acid and the evolved CO2 was detected by the carbon analyzer. The TOC results were compared with concentrations of organic materials determined from either ashing at 400°C for 6h (Jaffe and Walter, 1974) or ashing for l h at 500°C (Dean, 1974). The comparison indicated the former method is a satisfactory

171

6 8 r, o)

\

o 4 ~

~ e

o H

(D

g s (D

188 AMMONIA

CU 80

i~a A

~ - 6 8 / /

FE 48

2o

B 0 1 1

MN X 10 28 AMMONIA ~ 0 /

/ / /

I I

SODIUM HYDROXIDE

SODIUM HYDROXIDE

J /

{~.I i NORMALTTY OF EXTRACTANT

Fig. 1. Extraction of Cu, Zn, Fe, Mn and humic substances by ammonia and sodium hydroxide as a function of the strength of the extractant. Sediment from Tamar estuary.

substi tute for TOC analyses when an organic carbon analyzer is not available (Fig. 2). The latter method removes approximately 3 per cent of the sediment weight that is no t organic carbon.

Particle size. The proport ion of fine particles in the sieved sediments was estimated from the percent particles less than 14 ~um, as determined by the proport ion of the sediment mass which did no t settle after 15 min in an Andreasen's pipette apparatus.

STATISTICAL APPROACH

All data were transformed to logarithms for the statistical analyses. Linear regressions compared all combinations of extractable metals and extractable substrates. The unique nature of the data set (comparing a number of estuaries which vary widely in their physical and chemical characteristics) greatly helped circumvent the limitations of conducting statistical studies either over narrow data ranges or within single water bodies.

A Statistical filtering technique was designed to test ff substrates competed for the partitioning of a metal, and if that competi t ion was influenced by substrate concentrations in the sediment. It was assumed that, if two substrates were compet ing for binding a metal, a stronger statistical correlation between the metal and the first substrate would be observed among sediments

172

r~

z

W , <

t s l

Q

12

10

8

0 i i i i

0 I 2 3 4

ORGANIC CARBON (%)

/

i i

5 6

Fig. 2. A comparison of the concentration of organics determined by ashing at 400°C for 6 h (Jaffe and Walter, 1974) versus organic Carbon determined on a carbon analyzer. Solid line shows least squares best fit by equation y = 1.84 × + 0.484. Weight loss during ashing was corrected for weight loss due to salt in sediment.

with low concentrations of the competing substrate than would be observed if the competing substrate occurred in a range of concentrations in the data set. The statistical filter isolated subsets of data with low concentrations of one substrate. Correlations were then calculated between metals and all other substrates within that data set. If a metal and a substrate showed a progressively stronger correlation as the calculation was restricted to stations with progressively lower concentrations of the "compet ing" substrate, then it was concluded that compet i t ion was occurring between the two substrates. For example, if Fe and Mn oxides were competing for the binding of ametal , statistical correlations between Fe and the metal (signifying a partitioning of the metal to Fe oxide) would be weakened if stations occurred in the data set where the metal was predominantly part i t ioned to Mn. Since our objective is to test if the ability of Mn to compete with Fe for the metal was dependent upon the concentrat ion of Mn, we: (1)calculate the correlation between Fe and the metal among all stations; (2) eliminate those stations with very high Mn concentrations from the data set (this becomes subset 1) and recalcula te the Fe--metal correlation; (3)el iminate the stations with the highest Mn concentrations in subset 1, and recalculate the Fe--metal correlation in subset 2; and (4)progressively continue to eliminate high Mn stations until we reach a minimum of 12--16 data points (this resulted in 3-- 5 subsets, depending upon the range of data available for the substrate chosen as the competi tor , Table 2). If the correlation between Fe and metal

173

0

o

z ~

o

¢J

o

° ~

o ¢J

A

v

¢ q v

o

0 ~

o

~J

C)

~J

t~

174

improved progressively as we eliminated stations with high Mn from the data set, we concluded that Mn competed with Fe in a concentrat ion-dependent manner for partitioning of the metal. Furthermore, if the correlation between Mn and the same metal improved progressively as we eliminated stations with high Fe concentrations, the evidence for concentrat ion-dependent competi t ion between Fe and Mn was even stronger. By including different forms of substrates (as operationally-defined by the extracts) as either a correlating variable or a competing substrate, we could also test which operationally-defined forms were (statistically) most important in the partitioning process.

The statistical filter requires a large data base with a wide range of values. However, where adequate data is available, it provides an orderly, objective method for s tudy of interactions within specific subsets of data, which cannot be obtained from more conventional techniques (i.e., multiple regression). Because of the number of regressions calculated (> 15,000) only statistically significant results are reported here. Although filtering reduced the overall variance in the data set in some cases by progressively reducing the number of estuaries included in the calculation, the overwhelming majority of the correlations remained insignificant.

RESULTS

Sample collection Three lines of evidence indicated that sediments collected from the thin

surface layer on the mud flats were predominantly oxidized in character: (1) Surface sediments were obviously lighter in color than were subsurface sediments. (2) "In situ" measurements of Eh in light-colored surface sediments of estuarine mudflats consistently yield positive values (Luoma, unpublished data); and (3)Significantly higher concentrations of Cu, Pb, Zn and Co (Ag and Cd were no t measured) were extracted from the surface sediments than from the subsurface sediments by all extractants (exemplified by acetic acid extractions in Table 3). Metal sulfides should be of lower extractabili ty in most extractants than oxidized forms of sediment-bound metals. The difference between extractions from surface and subsurface were greater for Cu and Pb than for Zn and Co. These differences followed the solubility of the sulfides of the metals (Krauskopf, 1956), consistent with a predominantly oxidized form of metals in surface sediments and a domination of metal forms characteristic of a reducing environment in subsurface sediments. The following discussion relates only to extractions of the surficial oxidized sediments.

Su bstrate characterization Humic substances. Ammonia and NaOH removed similar quantities of

humic substances and trace metals. In the Looe estuary the humic substances extracted by ammonia const i tuted 23 percent of the total organic

175

TABLE 3

A COMPARISON OF Cu, Zn, Co AND Fe EXTRACTION FROM SURFACE SEDI- MENTS AND SUBSURFACE SEDIMENTS IN THREE ESTUARIES WITH VARYING METAL ENRICHMENT

Total metal Acetic acid metal

Surface Subsurface Surface sediment sediment sediment

Subsurface sediment

Copper Restronguet 3052 3648 1638 T a m ~ 282 306 95 Looe 36 49 5

Lead Restronguet 323 385 126 Tamar 179 222 97 Looe 78 117 48

Zinc Res~onguet 3544 3110 2627 901 T a m ~ 363 386 222 101 Looe 115 137 40 25

Cobalt Restronguet 23.2 27.3 12.6 8.0 Tamar 18.0 16.5 6.3 3.1 Looe 11.1 11.1 0.8 0.3

10 0.04 0.19

2.0 7.5 8

Iron Restronguet 48,383 46,531 10,802 12,096 Tamar 36,330 37,031 3,869 6,438 Looe 27,739 28,000 919 1,086

material in the sediments. If the calibration for the Looe sediments held throughout the study area, the humic materials varied from 1.0 per cent (Bristol Channel; Restronguet Creek) to 23 per cent (East Looe) of the organic matter among the different stations.

A single 1-week extraction did not remove all the humic substances from the sediments. Pretreatment of sediments with acid ammonium oxalate resulted in a substantial increase in the extraction of organic matter, Fe, and Cu by ammonia, compared to subsamples extracted without pretreatment (Table 4). In at least some sediments, a fraction of the humic substance is strongly enough associated with substances solubilized by acid ammonium oxalate (most likely Fe oxides -- Sholkovitz, 1976), to prevent total extraction of the humic substance by ammonia.

Fe oxides. A variety of extractants have been used to operationally characterize the concentrations, of different forms of Fe in sediments. In the sediments we considered, acid ammonium oxalate and 1N NC1 removed

176

T A B L E 4

A C O M P A R I S O N OF T H E E X T R A C T I O N OF HUMIC SUBSTANCES, Cu AND Fe BY A M M O N I A F R O M S E D I M E N T S P R E - T R E A T E D WITH A M M O N I U M O X A L A T E TO R E M O V E A M O R P H I C Fe, A N D F R O M U N T R E A T E D SEDIMENTS

S t a t i on C o n c e n t r a t i o n of h u m i c subs t ance (abs./g)

Ammonia~ex t r ac t ab l e me ta l conc.

Cu (pg/g) Fe (pg/g)

U n t r e a t e d Af t e r pre- U n t r e a t e d Af te r pre- U n t r e a t e d Af te r pre- t r e a t m e n t t r e a t m e n t t r e a t m e n t

Poole Harbor 6 .85 36.6 19.9 46.7 10 246 Beaul ieu R. 22.3 31.1 13.0 7.3 156 313 Looe I 4.8 11.3 4.9 5.4 26 149 Looe II 3.6 8.7 4.2 8.1 19 126

a similar proportion of the total Fe (Table 5). Acetic acid removed less than half of the quantity of Fe removed by oxalate. Earlier work using only sediments from the Tamar estuary (Bryan, unpublished data) indicated adding 0.1N hydroxylamine to 25 percent acetic acid (Chester and Hughes, 1967) extracted only slightly more Fe from oxidized sediments than did acetic acid alone (19.8 percent vs. 11.6 percent, respectively). Hydroxylamine alone (in nitric acid) extracted 7.6 percent of the total Fe from sediments where carbonates did not interfere with the extraction. The neutralization of the hydroxylamine extract by carbonates greatly reduced the extraction of Fe.

The differences among stations in the concentrations of extractable Fe were substantially greater than the interstation differences in total Fe. The wide variations in Fe extractability suggest processes independent of total concentration (such as those governing the crystallinity of the Fe oxide) play an important role in determining the quantity of Fe removed by each extractant. Laboratory experiments with pure Fe oxides also show that Fe extractability with acetic acid and with hydroxylamine decline with the age (and thus crystaUinity) of the precipitate (Luoma and Bryan, 1979). If crystallinity governs extractability, then increasingly crystalline forms of Fe are removed by extractants in the order oxalate = 1NHC1 > acetic acid + hydroxylamine ~ acetic acid ~ hydroxylamine.

The concentration of Fe removed by NaOH or ammonia correlated very strongly with the concentration of humic substance in the extracts (Fig. 3). Ammonia or NaOH may not quantitatively remove the organically-complexed Fe in the sediment (at high pH weakly compexed Fe may precipitate) but the concentrations which are removed by these techniques are controlled by the quantity of extractable organic materials in the sediment.

Mn oxides. The 10 different chemical treatments separated only 4 fractions of Mn: total Mn, a fraction of Mn (40--50 percent) similarly soluble in most acids and reducing agents, a smaller fraction (2.4--21.6 percent) soluble in

177

6 ,<

o

,< :>

m

E~

.<

E~

,<

,< [.., 0

0

0

.<

z o g

z

z © r ~

N

©

~ o I ~ 1 ° ~ I ~ I ~ I I ~ l ~

o 0

a0

T ~0

aO

I

0

I b-

"C-

o

0 0

g

0 0

o

~.o. @,1 T~

~ i ~ o ~ i ~

,o I

C)

J~ J~ T£ ~£ r f o 1 ~ o l m I o ~ , o

~ ~ ~ .

~ . o

. .

0

T

::L • .-~

: ~ <

0 0

0 ¢~ 0

0

• . .~ . . ~ ° °

"0 0

• 0~ 0 ~

178

0

t ~

• ° °

12',-

0

0 ~

o

I I cO D-- ~ c ~ e q

0

¢q

o ~. e,O ,,~ c ~ L O

I I ~ CO O0 c~

o;:

o

¢',1

Q;

eL

~ ~ N "~ 0

1000.

3.

° 100.0 C 0 E E o

10.0

c- O

O

C

c 0

(D

8.1

t+ .N,

t*

*#

t*

.0~

i I i i i i I i i i i i i i i i i i

1 0 . 0 I 0 0 .

C o n c e n t r a t i o n o £ Humic Subs tances

(obs /gm in ommonio )

179

1 0 . 0

v

o l . 80

o

z

"~ g . I

g 5

0.0

~e

I L L I I I I I I I I J I I J I

. 0 0 1 0 . 0 1 8 0 . 0

ConconkraLton of Humlc Subs tance~

(ab6/gm In ommonlo)

Fig. 3(a). Iron concentrat ions extracted by a m m o n i a as a funct ion of humic substance concentrat ions extracted by ammonia . (b). Manganese concentrat ions extracted by a m m o n i a as a funct ion of humic substance concentrat ions extracted by ammonia .

a m m o n i u m acetate, and a very small fract ion soluble in a m m o n i a or NaOH (Table 5). The fract ion o f Mn dissolved by weak acids correlated strongly with tota l Mn a m o n g all s tat ions (r = 0 .81 for oxalate to r = 0 .88 for acetic acid). The concentra t ion o f Mn extracted by a m m o n i u m acetate was

180

con t ro l l ed by processes largely i n d e p e n d e n t o f to ta l Mn c o n c e n t r a t i o n s (r = 0 .45 fo r co r re la t ion o f a m m o n i u m ace t a t e - -Mn and to ta l - -Mn) . Concen- t r a t ions in a m m o n i a and N a O H (as wi th Fe) were con t ro l l ed by the q u a n t i t y o f h u m i c subs t ance e x t r a c t e d (Fig. 3).

Extraction of trace metals H y d r o c h l o r i c acid r e m o v e d a grea ter c o n c e n t r a t i o n of t race me ta l f r o m

the sed imen t s t han o the r e x t r a c t a n t s (Table 5). Grea te r t han 60 pe r cen t o f the Ag, Cu, Pb and Zn in the ox id ized sed imen t s was r e m o v e d by HC1, a l t hough on ly 39 p e r c e n t o f the Co was HCl-soluble. Acet ic acid e x t r a c t e d subs tan t ia l ly m o r e Cu and Pb t h a n acid a m m o n i u m oxala te , bu t the two e x t r a c t a n t s r e m o v e d rough l y similar c o n c e n t r a t i o n s of Co and Zn. In the T a m a r sed imen t s (Bryan , unpub l i shed data) the m i x t u r e of h y d r o x y l a m i n e wi th acet ic acid resu l ted in an e x t r a c t i o n similar to acet ic acid a lone fo r Cu and Zn (34 vs. 44 p e r c e n t and 70 vs. 66 pe rcen t , respec t ive ly) , bu t e x t r a c t e d m o r e Pb than acet ic acid (68 vs. 42 pe rcen t ) . C o n c e n t r a t i o n s of m o s t meta l s in the th ree acids were s t rong ly co r re l a t ed wi th the to t a l me ta l c o n c e n t r a t i o n s in the s e d i m e n t (Table 6), indica t ing ex t r ac t ab i l i t y was largely d e p e n d e n t u p o n to ta l c o n c e n t r a t i o n .

A high p r o p o r t i o n , re lat ive to o the r meta ls , o f t o t a l Cu and Ag was re- m o v e d f r o m the sed imen t s by N a O H and a m m o n i a (Table 5), ind ica t ing the relat ive i m p o r t a n c e o f humic subs tances as a subs t ra te fo r b inding Cu and Ag. In te res t ing ly , the sum of acet ic ac id - -Cu and a m m o n i a - - C u was similar to the c o n c e n t r a t i o n s of Cu e x t r a c t e d by HC1 a t m o s t s ta t ions (Table 7). S t rong devia t ions f r o m this re la t ionsh ip were observed on ly a t 2 s ta t ions

TABLE 6 CORRELATION OF EXTRACTABLE METAL CONCENTRATION WITH TOTAL METAL CONCENTRATION r -- Correlation coefficient, b -- Slope of regression, n -- 45 for all except pyrophosphate where n -- 17 and Ag--ammonia, were n --- 21. a = Insignificant correlation, p > 0.05. nd -- Not determined

Extractant Ag Co Cu Fe Mn Pb Zn

Hydrochloric acid

Acetic acid

Oxalate

Ammonium acetate

P y r o p h o s p h a t e r b

r 0.70 0.77 0.88 0.62 0.84 1.00 1.00 b 0.53 1.66 1.08 1.38 1.39 1.06 1.16

r nd 0.81 1.00 0.67 0.88 1.00 1.00 b 2.04 1.11 1.61 1.43 1.12 1.21

r nd 0.86 0.52 0.69 0.81 0.69 1.00 b 1.68 0.80 1.68 1.35 1.04 1.24

r nd 0.33 0.55 0.40 0.45 0.62 0.90 b 0.86 0.81 2.08 0.87 1.09 1.49

nd 0.11 a 0.90 0.22 a 0.48 0.67 0.95 0.23 0.96 0.41 0.89 0.82 1.17

181

T A B L E 7

A C O M P A R I S O N O F HCI--Cu WITH T H E SUM O F (ACETIC ACID- -Cu) + (AMMONIA - -Cu) A T S E L E C T E D S T A T I O N S A N D A L O N G THE S A L I N I T Y G R A D I E N T OF 4 E S T U A R I E S

Estuary A B ( A ) + ( B ) HCI--Cu ( u g l g ) (ug/g)

Acetic acid---Cu Ammonia--Cu (Pg/g) (Pg/g)

Plym 25.1 23.7 48.8 49.1 Erme 9.3 8.2 17.5 14.5 W. Looe 13.7 12.6 26.3 27.8 Fal (mouth) 1423 367 1790 1770

1638 399 2037 1823 (head) 426 240 666 1860 a

Tamar (mouth) 154 59 213 194 95 50 145 168

154 55 209 233 127 53 180 253 165 60 225 260 149 47 196 250

(head) 700 237 937 605 a F o w e y ( m o u t h ) 26 13 39 49

22 15 37 44 23 17 37 50 33 20 53 50 35 24 59 65

(head) 11 18 29 55 a Gannel (mouth) 28 11 39 41

• 35 11 46 41 71 30 101 87

(head) 135 49 184 197 Beaulieu 10 7 17 12 Poole 16 23 39 90 a Hayle 493 121 614 645 Bristol Channel 17 5 22 23

astations with substantial deviations between HC1--Cu and acetic acid Cu + ammonia-- Cu.

where the samples included some reduced sediments due to sampling diffi- culties (thus inhibiting the extraction of Cu by acetic acid) and three stations near the freshwater--seawater interface. Excluding the anomalous stations the regression of HC1--Cu vs. (acetic acid--Cu) + (ammonia--Cu) had a correlation coefficient of 0.98, a slope of 0 .98 and an intercept of 1.07 #gig Cu. These results suggested ammonia extracted roughly the same organically complexed Cu as HC1, but that this form of Cu was not extracted by acetic acid. Extraction schemes which employ acetic acid to determine "non- detrital" trace metals (Loring, 1976) may miss at least some organically complexed forms of Cu.

The concentrations of Ag, Cu and Zn extracted by the alkaline extractants

1 8 2

r ~

D Z

z o

Z o

.<

0 z o

Z ~

o ~

O ~

0 ~

og

~ 0

°~

o D

e~

o

0

% %

I

I

~.~ 0'3 b - ¢.0 o ¢q 0 o o c; o

I

o. o o o 0 0 0 0

I

0 0 0 0 0 0 ¢~

I

0 0 0 0 0 0

I

0 0 0 0 0 ¢~ 0

g

0 o. 0

V

o 0

V

0 0 0

V

183

T A B L E 9

C O R R E L A T I O N O F E X T R A C T A B L E F R A C T I O N S O F Co, Zn , Pb, Cu, Ag A N D Cd WITH O R G A N I C C A R B O N A N D S E L E C T E D F R A C T I O N S OF Fe A N D Mn N u m b e r s are co r r e l a t i on coef f ic ien ts (r). E x t r a c t a n t s s h o w n are those wh ich gave s t ronges t cor re la t ions . Cor re l a t ions w i th oxa l a t e - -Fe are s h o w n in Tables 12--15.

Metal MnAmAc Mnoxal" Mntota 1 FeHC 1 Fetota ! Organic E x t r a c t i o n c a r b o n

Co Acet ic acid 0 .42 a 0 .63 b 0 .55 b 0 .46 b 0 .56 b 0 .12 IICl 0 .40 a 0 .675 0 .60 b 0 .48 b 0 .43 a 0 .23 Oxa la te 0 .43 a 0 .75 b 0 .75 b 0 .545 0.51 b 0 .18 Tota l 0 .41 a 0.71 b 0 .75 b 0 .58 b 0 .70 b 0 .19

Ztl A m m o n i u m

ace ta t e 0 .30 c 0.21 0 .38 d 0 .64 b 0.41 a 0 .19 Ace t ic acid 0 .30 c 0 .15 0 .36 d 0 .79 b 0 .52 b 0.41 a HCI 0 .25 0 .10 0 .35 c 0 .82 b 0 .52 b 0 .44 a Oxa la te 0 .27 0 .15 0 .38 d 0 .80 b 0 .50 b 0 .42 a Tota l 0 .26 0 .14 0.37 d 0 .795 0 .58 b 0 .36 c

Pb Acet ic acid 0 .35 c 0 .31 c 0.37 d 0 .35 d 0.37 "c 0 .03 HC1 0 .36 c 0 .33 c 0 .40 d 0 .46 b 0 .42 a 0.07 Oxa la te 0 .32 c 0 .35 c 0 .22 0 .30 a 0.27 0 .06 To ta l 0 .38 d 0 .31 c 0 .30 c 0 .48 b 0 .40 d 0 .05

Cu Acet ic acid 0 .36 c 0 .40 a 0 .41 a 0 .49 b 0 .41 a 0 .13 HC1 0 .28 c 0 .13 0.27 0.71 b 0 .54 b 0 .39 a Oxa la te 0 .25 0 .595 0 .50 b 0 .28 c 0 .29 ( - - )0 .10 Tota l 0 .29 c 0 .10 0 .30 c 0 .69 b 0 .52 b 0.33c

Ag HC1 0.01 - - 0 . 1 2 0.17 0 .29 c 0 .49 b - -0 .07 Tota l 0 .12 - - 0 . 0 5 0 .09 0.57 b 0 .47 b 0.21

Cd HC1 0.10 - -0 .21 - -0 .21 0 .50 b 0 .15 0.49 b

ap ~ 0 .005. bp ~ 0 .001. Cp ~ 0.05. dp ~ 0.01.

w e r e s t r o n g l y c o r r e l a t e d w i t h t o t a l A g , C u a n d Z n i n t h e s e d i m e n t s , b u t n o t

w i t h t h e a b u n d a n c e o f h u m i c s u b s t a n c e s .

Statistical analyses Substrates. A m o n g s u b s t r a t e s , t h e s t r o n g e s t c o r r e l a t i o n s w e r e o b s e r v e d

b e t w e e n t o t a l o r g a n i c c a r b o n a n d e x t r a c t a b l e F e ( T a b l e 6 ) . T h i s r e l a t i o n s h i p

w a s a t l e a s t p a r t l y d u e t o a s i m i l a r d e p e n d e n c e o f e x t r a c t a b l e F e a n d o r g a n i c

1 8 4 .

18~2

r~

H

rv

bJ

L)

ALL DATA

i1.

b ii.

t,l. i~, w.

w ~

, 8 I I I I I I I I I

18

CONCENTRATION OF TOTAL FE ( ~ : ;

(a)

S T A T I O N S WHERE HN CONCENTRAT IONS LESS THAN 4 8 8 u g / g

l e 8

F-

z o

c~

z w

c~ c3

Fig. 4 .

m

W ~" K.

N ~ ii.iI

w. ii. i(, tl. I,(.

w.l~.

NWw. WI.

. 0 I I I I I I I I I

C O N C E N T R A T I O N OF TOTAL FE ( % )

(b )

185

STATZONS WHERE MN CONCENTRATZONS LESS THAN 2@8 ug/g

b

H

0~

[ d (D

D

B

D

m

N, 14

14

. 0

I I I t I t I I, I IO

CONCENTRATTOM O F TOTAL F E ( Y ~ )

(c)

Fig. 4. Correlation of total Co in sediment with total Fe in sediment (a) among all stations; (b )among stations with concentrations of extractable Mn less than 400pg/g; and (c) among stations where concentrations of extractable Mn were less than 200 pg]g.

materials upon particle size distributions, since concentrations of both correlated significantly with the proportion of fine particles in the sediments. The significant, negative relationship between carbonates and humic substances reflected physical rather than chemical processes. Humic substances declined in concentration from the head to the mouth of most estuaries and occurred in low concentrations at stations heavily influenced by marine processes. Carbonates occurred in high concentrations where marine influences were greatest, and increased from the head to the mouth of most estuaries.

Cobalt. Among all samples, Co correlated most strongly with total Fe, total Mn and oxalate-soluble Mn (Table 9). A multiple regression calculation indicated total Fe and total Mn together explained 88 percent of the variance in concentrations of Co in the sediments.

Out statistical filtering method was used to test whether Fe and Mn competed for the binding of Co at different stations. The correlation between total Fe and total Co became progressively stronger when stations with high Mn concentrations were eliminated from the calculation (Fig. 4). Thus, stations where concentrations of Mn were high (~300 ttg/lg) tended not to fit the Fe--Co relationship, indicating a partitioning of Co away from Fe where competition from Mn might be most effective. Similarly, the correlation of Co and Mn was strongest among subsets of samples with low concentrations of Fe (Table 10). Stations with high concentrations of Fe did not fit well in the Co--Mn relationship. These results suggest Co was

186

TABLE i0

CORRELATION OF Co WITH Fe AND Mn WITHIN SUBSETS OF DATA WHICH EXCLUDE THE SAMPLES FROM THE PREVIOUS SUBSET WITH THE HIGHEST CONCENTRATIONS OF THE SPECIFIED COMPETING SUBSTRATES. a

Filter Correlation of coefficients (r)

Competing Fraction All Subsetl Subset2 Subset3 Subset4 Subsets substrate samples

Fetota 1 vs COtotal b

Mn acetic 0.71 c 0.81 d 0.83 d 0.97 d 0.99 d 1.00 d acid

Mntota 1 vs COtotal b

Fe oxalate 0.75 d 0.80 d 0.98 d 0.97 d Fe total 0.75 d 0.82 d 0.82 d 0.90 d

a Filter levels and sample sizes for each subset of data are shown in Table 2. bSimilar results were obtained with all forms of acid extractable Co. Cp < 0.005. dp ~ 0.001.

dominantly parti t ioned to Mn oxides where concentrations of Fe oxide were low, and where concentrations of Mn oxide were low, Co was dominantly parti t ioned to Fe oxide.

Zinc. Both acid-soluble and total Zn correlated very strongly with concentrations of Fe in the sediments, and correlations with extractable Fe were substantially stronger than correlations with total Fe (Table 9). A very strong correlation of Zn with Fe occurred among stations low in Mn, total organic carbon or humic substances (Table 11), suggesting partitioning of Zn away from Fe where these substrat~s were in high concentration. More- over, Zn correlated only weakly with organic materials and with Mn among all samples; but where Fe or Fe/Mn were low, both correlations were greatly improved.

If the statistical relationships result from the mechanism we have postulated, then extractable forms of Fe dominated the partitioning of Zn in most sediments; but where Fe concentrations were low and/or the concentrations of either Mn or organic materials were high, the latter two substrates were also important in binding Zn.

Lead. Concentrations of Pb in sediments correlated weakly, but significantly, with Fe and Mn among all samples (Table 9). Similar to Zn, a very strong correlation of Pb with extractable-Fe (but no t total Fe) was observed among stations where either Mn or organic carbon were low in concentration (Table 12). A strong correlation of Pb with Mn was also observed where Fe/ Mn was low, suggesting strong competi ton between at least extractable Fe and Mn for the binding of Pb in the sediments. The correlation of Pb with

187

TABLE 11

C O R R E L A T I O N OF Zn WITH Mn, ORGANIC CARBON AND E X T R A C T A B L E Fe WITHIN SUBSETS OF D A T A WHICH EX CL U D E THE SAMPLES F R O M THE PRE- VIOUS SUBSET WITH THE HIGHEST CONCENTRATIONS OF THE SPECIFIED COMPETING SUBSTRATES a

Fil ter Corre la t ion coeff ic ients (r)

Competing Fraction All Subsetl Subset2 Subsets Subset4 Subsets substrate samples

ZnHC1 b vs. Feoxal

Mn acetic 0.82 c 0.84 c 0.85 c 0.91 c 0.94 c 0.92 c acid

Organic ca rbon 0.82 c 0.81 c 0.80 c 0.83 c 0.95 c Humic subs tances 0.82 c 0.85 c 0.89 c 0.93 ¢ 0.91 c

ZnHcl b vs. Organic ca rbon

Fe oxala te 0.44 d 0.47 d 0.33 0.81 ¢ Mn total 0.44 d 0.47 d 0.77 c 0.92 c Mn acetic 0.44 d 0.50 d 0.53 c 0.63 ¢

acid

ZnHC1 b vs. Mntotal

Fe /Mn 0.35 e 0.49 d 0.74 c 0.84 c 0.92 ¢

0.60 c 0.79 c

a Fi l ter levels and sample sizes for each subse t o f data are shown in Table 2. bSimilar results ob ta ined wi th tota l Zn. Cp 0.001. dp 0.005. ep 0.05.

organic carbon where Mn was low may reflect either a role for organic materials in Pb partitioning which was not statistically evident among all samples; or a secondary effect of the correlation between organic carbon and extractable Fe.

Copper. Statistical correlations among all samples were consistent with previous conclusions that HC1 extracted both organically and inorganically- bound forms of Cu, but acetic acid only extracted Cu from inorganic substrates (Table 9). Extractable Fe also showed a stronger relationship with Cu than did total Fe.

A very strong relationship between Cu and extractable Fe was found where humic substances, Mn or total organic carbon were low in concentration (Table 13). The deviations from the Fe--Cu relationship which occurred at high humic substance concentrations, and the high proportion of Cu soluble in ammonia, both indicated the importance of humic substances in the partitioning of Cu. The correlation of Cu with Mn, organic

188

T A B L E 12

C O R R E L A T I O N O F Pb WITH Mn, O R G A N I C C A R B O N A N D E X T R A C T A B L E Fe WITHIN SUBSETS OF D A T A WHICH E X C L U D E THE SAMPLES IN THE P R E V I O U S S U B S E T WITH THE H I G H E S T C O N C E N T R A T I O N S OF THE S P E C I F I E D COMPET- ING S U B S T R A T E S a

F i l te r Cor re la t ion coef f ic ien ts (r)

Substrate Fraction All Subsetl Subset2 Subset 3 Subset4 Subsets samples

Pbtotal b vs. Feoxal

Mn acet ic acid c 0 .48 d 0 .56 d 0 .56 d 0.91 d 0 .99 d

Organic c a r b o n 0.48 d 0 .46 e 0 .46 f 0 .48 f 0.91 d

Pbtotal b vs. Mntota I

F e / M n acet ic acid 0 .38 f 0 .43 f 0 .50 e 0 .76 d 0 .75 e

Pbtotal b vs. organic c a r b o n

Mn to ta l 0 .05 0 .19 0.64 e 0 .78 d

1 .00 d

aF i l t e r levels and sample sizes for each subse t of da ta are s h o w n in Table 2. bSl igh t ly weaker b u t s imilar responses were observed wi th HC1--soluble Pb. c A sl ight ly weaker b u t s imilar r e sponse was observed to the to ta l Mn fil ter. ap < 0.001. e p ~ 0 .005. f p ~ 0.01.

carbon or humic substances was not progressively or substantially improved where Fe concentrations were low. Extractable forms of Fe appeared to dominate the partitioning of Cu, but the relationship of Cu with the other substrates was difficult to clearly resolve from statistical comparisons.

Silver. Hydrochloric acid extracts Ag as a soluble Ag--chloro complex. HC1--Ag correlated relatively weakly with total Ag(r = 0.70). The proportion of Ag solubilized by HC1 was lowest in sediments characterized by either high concentrations of Fe (Fal, Tamar estuaries) or high concentrations of Mn (Gannel estuary). HC1--Ag also correlated more weakly with extractable- Fe than did total Ag (Table 9); and the improvement in the correlation of extractable Fe and Ag which occurred among stations with low Mn concentrations was greater for total than for HCl-soluble Ag (Table 14). Both fractions of Ag showed a similar correlation with Fe where humic substances were low, however. The correlations suggest (1) Ag is primarily partitioned among extractable forms of Fe, humic substances (as also suggested by NaOH and ammonia extractions) and perhaps Mn oxides in oxidized

189

TABLE 13

C O R R E L A T I O N OF Cu WITH E X T R A C T A B L E Fe WITHIN SUBSETS OF DATA WHICH E X C L U D E THE SAMPLES F R O M THE P R E V I O U S SUBSET WITH THE HIGHEST CONCENTRATIONS OF THE SPECIFIED COMPETING SUBSTRATES a

Fil ters Corre la t ion Coeff ic ients (r)

Competing Fraction All Subset1 Subset2 Subset3 Subseta Subsets substrate samples

Cuto ta l b vs. Feox ~

Humic subs tances 0.69 c 0.78 c 0.84 c 0.95 c 0.96 c Mn acetic

acid 0.69 c 0.71 c 0.72 c 0.81 c 0.89 c Organic ca rbon 0.69 c 0.72 c 0.71 c 0.75 c 0.80 c

0.77 c

aFi l te r levels and sample sizes for each subset o f data are shown in Table 2. bHCl-soluble C u shows a slightly weaker bu t similar t rend . Cp < 0.001.

TABLE 14

C O R R E L A T I O N OF Ag WITH E X T R A C T A B L E Fe AND HUMIC SUBSTANCES WITH- IN SUBSETS OF DATA WHICH E X C L U D E THE SAMPLES FROM THE PRE V IO U S SUBSET WITH THE HIGHEST CONCENTRATIONS OF THE SPECIFIED COMPET- ING SUBSTRATES a

Fil ter Corre la t ion coeff ic ients (r)

Competing Fraction All Subsetl Subset2 Subset 3 Subset4 Subset s substrate samples

AgHC 1 vs. Feoxa]

Humic subs tances 0.31 b 0.51 c 0.63 d 0.67 d 0.77 c Mn acetic

acid 0.31 b 0.31 b 0.35 b 0.43 b 0.54 c

Agtota I vs. Feoxal

Humic subs tances 0.58 d 0.67 d 0.74 d 0.79 d 0.75 c Mn acetic

acid 0.58 d 0.58 d 0.60 d 0.73 d 0.84 d

Agtotal vs. humic subs tance

Fe oxalate 0.08 0.16 0.37 0.57 b

0.50 b

0.94 d

aF i l t e r levels and sample sizes for each subse t o f data are s h o w n in Table 2. bp < 0.05. Cp < 0.005. dp < 0.001.

190

T A B L E 15

C O R R E L A T I O N O F Cd WITH O R G A N I C C A R B O N AND E X T R A C T A B L E Fe WITHIN SUBSETS O F D A T A WHICH E X C L U D E T H E SAMPLES F R O M THE P R E V I O U S SUB- SET WITH T H E H I G H E S T C O N C E N T R A T I O N S OF THE SPECIFIED C O M P E T I N G S U B S T R A T E S a

F i l t e r Cor re l a t ion Coeff ic ien ts (r)

Competing All Subsetl Subset2 Subset3 Subset4 Subsets substrate samples

CdHCl vs. organic c a r b o n

Mn acet ic acid 0 .49 b 0.51 b 0 .58 b 0.68 b 0 .67 b

Humic subs tances 0 .49 b 0 .50 c 0 .44 d 0 .30 0.09

CdHc I vs. Feoxal

Mn acet ic acid 0 .49 b 0 .47 c 0 .47 c 0 .60 b 0 .63 b

Organic c a r b o n 0 .49 b 0 .45 c 0 .41 d 0.51 e 0 .65 d

0 .68 b

0 .66 b

a F i l t e r levels and sample sizes for each subse t of da ta are s h o w n in Table 2. b p < 0 .001.

dp < 0 .005 . p < 0.05.

e p < 0 .01.

sediments; and (2)HC1 extracts Ag bound to humic substances, bu t does no t extract a proport ion of the Ag associated with Fe and/or Mn oxides.

Cadmium. Among all samples, Cd correlated significantly with amorphous Fe, organic carbon and particle size (Table 9); and among sediments low in Mn the correlations of Cd with all three variables were enhanced (Table 15). However, amorphous Fe, organic carbon and particle size are intercorrelated (Table 8), thus the relative importance of the three variables in Cd partitioning could no t be separated.

Manganese. Manganese may occur in oxidized sediments as Mn carbonate, Mn oxide or Mn bound to other substrates. Strong correlations between Mn and carbonate occurred in sediments low in either organic materials or total Fe (Table 16), suggesting Mn carbonate was more common where substrates which bind Mn were in low concentration. Similarly, the strong correlation of Mn with organic materials in sediment low in carbonates indicated complexation of Mn was important where Mn carbonate was least likely to form in abundance. Manganese also correlated significantly with particle size (increasing in concentrat ion in finer ga ined sediments) in low carbonate sediments. Such a correlation could be a secondary result of complexation with organic materials in environments where Mn carbonate was less abun- dant, since Mn oxides tend to occur as discrete particles (Jenne, 1968; 1977).

191

T A B L E 16

C O R R E L A T I O N O F Mn WITH C A R B O N A T E , P A R T I C L E SIZE A N D O R G A N I C C A R B O N WITHIN SU B SE T S O F D A T A WHICH E X C L U D E THE SAMPLES F R O M THE P R E V I O U S S U B S E T WITH T H E H I G H E S T C O N C E N T R A T I O N S O F THE SPECI- F I E D C O M P E T I N G S U B S T R A T E S a

Filters Correlation of coefficients (r)

Competing Fraction All Subsetl Subset2 Subset 3 Subset4 substrate samples

MnHCI b vs. carbonate

Fe To ta l 0 .15 0 .34 c 0 .67 d 0 .72 d Total organic carbon 0.15 0 .35 c 0 .37 ¢ 0 .59 d 0.81 d

MnHC1 b vs. particle size

Carbonate 0.06 0 .14 0 .18 0 .34 0 .52 c

MnHcl b vs. TOC

Carbonate 0.09 0 .18 0.27 0 .49 e 0 .62 f

aFilter levels and sample sizes for each subset of data are shown in Table 2. bSimilar effects, but weaker correlations were observed with total and oxalate-Mn. Cp < 0.05. d ~ 0 .001 . ep < 0.01. fp < 0 .005.

Acidic extractions (HC1, acetic acid, acid ammonium oxalate, hydroxylamine--nitric acid) appeared to solubilize all three forms of Mn, since no correlation of Mn with any substrate was evident in acid extractants when all samples were considered.

Evidence from calculations with Co, Zn, Pb, Cu and Cd indicated ammonium acetate selectively extracted Mn oxide in low carbon sediments. Corre- lations of all five metals with ammonium acetate--Mn were weak among all sediments but very strong among sediments low in organic carbon. Hydrous oxide is the form of Mn which binds metals most strongly in sediments. Where little Mn was expected in complexed form (sediments low in organic carbon) the association of metals with ammonium acetate-soluble Mn was strong. Metal correlations with acid-soluble Mn also improved slightly in low carbon sediments (since less organically complexed Mn would be extracted from such sediments the association of the trace metals with the form of Mn involved in binding may have been more evident), but the improvements were not as dramatic as with ammonium acetate--Mn (since the acids extract both Mn carbonate and Mn oxide).

Common effects among metals Several observations c o m m o n among a number of metals were also evident

192

in the calculations. No metal except Cd correlated significantly with the proport ion of fine particles in the sediments, either among all samples or in any subset of samples. Particle size is well established as an important variable affecting trace metal concentrations within samples or over narrow concentrat ion ranges (Bryan and Hummerstone, 1977). Through a wide data range, however, anthropogenic/geologic enrichment and chemical associations may swamp out statistically detectable particle size effects among samples.

All metals except Co, correlated negatively with carbonates in sediments which were low in humic substances (i.e., sediments most influence by marine processes). Such correlations could reflect a substantial dilution of metal concentrations where carbonates occur in abundance, or a net de- sorption of metals with increasing marine influence in seaward estuarine reaches. The former explanation seems most likely, since calcium carbonates const i tuted up to 30 percent of the sediment weight in some areas. If so, comparisons of trace metal enrichments within and among estuarine stations should be calculated on a carbonate-free basis.

DISCUSSION

Extracting oxidized sediments Chemical characterization of the sediments in estuaries will be grossly

affected by procedures used to collect and store sediments, prior to extraction. Since the oxidized sediments in most estuaries lie in a thin layer at the sediment--water interface, samples collected by procedures (grabs, mixed cores, etc.) which do no t include subsurface sediments may show substantially different extractabili ty of metals and substrates than would be observed if only surface sediments were collected. Storage procedures which allow sediments to become anoxic will yield similar results (Thomson et al., in press). A comparison with our results indicate that most studies which have a t tempted, to date, to characterize sediment-bound metal forms with chemical extractants have considered anoxic sediments (Gupta and Chen, 1976; Serne, 1977; Brannon et al., 1976), despite the chemical and biological importance of the predominantly oxidized sediments at the sediment--water interface.

The extractions were quite useful in describing the fractions of substrates which reacted most strongly with trace metals. Both direct extractions and statistical relationships indicated that the alkali.soluble fraction of the organic substrate was very important in the organic complexat ion of at least Cu and Ag. The extractable fraction of Fe in the sediment showed a stronger association with Ag, Cd, Cu, Pb and Zn (but no t Co) than did total Fe, reflecting the higher density of sorption sites in the amorphous Fe phases removed by extractants (Jenne, 1977). The fraction of Fe extracted by acid ammonium oxalate or 1N HC1 correlated more strongly with trace metals than did the smaller quanti ty of Fe removed by acetic acid. The low solubility in acid extracts of both Co and Fe, substantiated the statistical inference

193

that Co associated with a more crystalline fraction of Fe than did other metals. The association of Co with total Fe indicated either: (1)specif ic sorption sites of low density, bu t with a high affinity for Co, may occur in crystalline Fe forms; or (2) much of the Co in these estuaries originated in geologic association with some crystalline Fe mineral. The latter would suggest that nearly all the non-detrial Co was parti t ioned to Mn.

The extractants operationally defined 3 fractions of Mn, but did not clearly resolve the partitioning of Mn in the sediments. Among all stations, Co, Zn, Pb and Cu generally correlated more strongly with total Mn than with any of the extractable fractions. If we assume Mn oxides are more important in metal partitioning than Mn carbonates or sorbed Mn, the strong correlations of metals with total Mn suggested most extractants contain propor t ionatey more Mn carbonate or bound Mn, and less Mn oxide than does the concentrated nitric acid digest (i.e., much of the trace metal- reactive Mn oxide in sediments is no t solubilized by weak acid extraction). The presence of Mn carbonates in the sediments is evidence by the correlations of Mn and carbonate. However, the improvement of correlations between metals and extractable Mn at stations where concentrations of organic carbon were low rather than in sediments low in carbonates, suggested complexed forms of Mn made-up a larger fraction of the Mn extracted by acids than did Mn carbonates.

The extractions were of minimal value in directly removing specific metal forms, as expected (Guy et al., 1977; Luoma and Jenne, 1976). Concentra- tions of acid-soluble metal correlated strongly with total metal in most instances; thus the fraction of metal involved in partitioning could no t be determined from the statistical relationships within the extraction data. The strongest metal--substrate correlations were observed with either HC1- soluble or total metal, suggesting HC1 would be the preferred method for an operational definition of "ext rac table" metal concentration in sediments (as also concluded by Malo, 1977, and Agemian and Chau, 1977). Eaton (1978) recently employed acid ammonium oxalate to assess extractable metal in San Francisco Bay sediments. This method extracted low and highly variable quantities of Pb and Cu from the sediment we studied. Either de- sorption of Cu and Pb is ineffective at pH 3.3, or some resorption of metals to undissolved substrates (or precipitation in the case of Pb) occurs during the extraction. Acetic acid has also been used to describe extractable and biologically available metals (Loring, 1976); however, in some sediments acetic acid does no t extract organically complexed Cu removed by HC1.

The fraction of metal removed from sediments by most weak acid extractants was a constant function of the total metal in the sediment. Although the propor t ion of weak acid--soluble metal increased with enrichment (except for Ag), the correlation with total metal suggested no acid extractant success- fully removed all the non-lithogenic metal. The formation of sedimentary particles by successive layering of adsorption substrates (Jenne, 1977), the solid state diffusion of metals into the interstices of substrates such as manganese oxides (Jenne, 1977), and the movemen t of metals into lattices

194

after adsorption (Farrah and Picketing, 1978) may limit the efficiency of acid extractions. A single extraction with a weak acid may aid in assessing the biological availability of some metals from sediments (Luoma and Bryan, 1978; 1979) but our data suggest no extractant will satisfactorily define the anthrophogenically-derived metal in a sediment (as proposed, for example, by Agemian and Chau, 1977).

Useful information on metal partitioning in oxidized sediments may be optimally obtained from simple partial extraction schemes which:

(1) Employ HC1 of sufficient acidity to buffer carbonate neutralization; (2) Employ an alkaline extractant to characterize concentrations of humic

substances; and (3) Characterize extractable Fe and Mn oxides with acid ammonium

oxalate. Correlations suggest IN HC1 may be employed in place of acid ammonium oxalate with only a small loss of information. An additional extraction with ammonium acetate could be added to gain information about partitioning to Mn oxides.

The information obtained from such extractions may be useful in statistically illustrating metal--substrate associations, defining the relative abundance of different substrate forms for use in partitioning or bioavailability models, or empirically defining a less-than-total, "extractable phase" of metals in sediments.

Mn

Mfl o.,,o..,. I I I

"Operationally-defined

Fig. 5. A flow diagram of metal partitioning in oxidized sediments. The width of the arrows roughly signifies partitioning in most sediments in south and west England estuaries; however, in any given sediment partitioning may vary with concentrations of substrates.

195

Partitioning Statistical evidence indicated Ag, Cd, Co, Cu, Pb and Zn were all partitioned

among a variety of competing substrates in these oxidized estuarine sediments (Fig. 5). The relationships indicated partitioning was characterized by competion between substrates. The outcome of the competition was influenced by substrate concentrations. Hydrous oxides of Fe appeared to dominate the partitioning of Zn, Pb and, possibly, Cd in most sediments; oxides of Fe and humic substances were both generally important in the partitioning of Cu and Ag; and Co appeared to be equally partitioned between hydrous oxides of Fe and Mn. However, organic materials and oxides of Mn were also important in partitioning where sediments contained low concentrations of Fe, except in the case of Co.

Generalizations which infer metal associations with one type of substrate may be overly simplistic. Metal partitioning in sediments is a dynamic process, subject to influence by the physicochemical environment. Metals may be partitioned among several forms simultaneously. The dominant form of any metal may differ substantially among chemically different sediments or within sediments over time.

We have used statistical relationships to test the postulated existence of mechanisms such as concentration~lependent competition among sub- strated for metal partitioning. Statistical relationships do not prove cause and effect, nor does the existence of correlations predictecl from our hypo- theses irrefutably prove the postulated mechanisms exist. A statistical approach may be necessary to help identify the important variables in complex natural systems, however, and may be most useful in pointing to specific studies which more directly test the existence of such mechanisms.

ACKNOWLEDGEMENTS

We would like to thank Mr. G. Burt and Mr. L.G. Hummerstone for their assistance with sampling and analysis, and Mr. L. Barker for his assistence with the computer work. The study was partly supported by United Kingdom Department of the Environment under contract DGR 480/51, a U.S. Na- tional Science Foundation NATO Postdoctoral Fellowship and U.S. Environ- mental Protection Agency passthrough funds. We also appreciate the valuable comments of John Hem and Lorraine Filipek of the U.S.G.S.

REFERENCES

Agemian, H. and A.S.Y. Chau, Arch. Eviron. Contamination and Toxicology, 6 (1977) 69.

Brannon, J.M., R.M. Engler, J.R. Rose, P.G. Hunt and I. Smith, U.S. NTIS, AD Rep AD a026355, 1976, p. 44.

Bryan, G.W. and L.G. Hummersone, J. Mar. Biol. Assoc. U.K., 51 (1971) 845. Bryan, G.W. and H. Uysal, J. Mar. Biol. Assoc. U.K. 58, (1978) 89.

196

Chao, T. T., Proc. Soil Sci. Soc. Am., 31 (1972) 764. Conomos, T. J. and D. H. Peterson, in: Estuarine Processes--Circulation, Sediments and

Transfer of Material in the Estuary, Vol. 2, Academic Press, New York, 1977, p. 82. Dean, W. E., J. Sediment. Petrol., 44 (1974) 242. DeGroot, A. J., W. Solomons and E. Allesma, in: Estuarine Chemistry, Academic Press,

London, 1976, p. 131. Eaton, A., Environ. Geol., 2 (1978/1979) 333. Farrah, H. and W. F. Pickering, Water, Air, Soil Pollut., 9 (1978) 23. Gupta, S. K. and K. P. Chen, Environ. Lett., 10 (1975) 129. Guy, P. D., C. L. Chakraborti and D. C. McBain, Water Res., 12 (1978) 21. Hem, J. D., Geochim. Cosmochim. Acta, 41 (1977) 527. Hem, J. D., Chem Geol., 21 (1978) 199. Jaffe, D. and J. K. Walters, Sci. Total Environ., 7 (1977) 1. Jenne, E. A., in: Molybdenum in the Environment, Vol. 2. Dekker, New York, 1977, p.

125. Krauskopf, K. B., Geochim. Cosmochim, Acta, 9 (1956) 1. Loring, W. H., Can. J. Earth Sci., 13 (1976) 960. Luoma, S. N. and G. W. Bryan, J. Mar. Biol. Assoc. U.K., 68 (1978) 793. Luoma, S. N. and G. W. Bryan, in: Chemical Modelling -- Speciation, Sorption, Solubility,

and Kinetics in Aqueous Systems. American Chemical Society, Washington, DC, 1979, p. 577.

Luoma, S. N. and E. A. Jenne, in: Trace Substances in Environmental Health. X. Univer- sity of Missouri Press, Columbia, 1976, p. 343.

Luoma, S. N. and E. A. Jenne, in: Biological implications of Metals in the Environment. Avail. as CONF-750929, U.S. NTIS, Springfield, VA, 1977, p. 213.

Malo, B. A., Environ. Sci. Technol., 1 (1977) 277. Morel, F. M. M., J. C. WestaU, C. R. O~Melia, and J. J. Morgan, Environ. Sci. Technol., 9

(1975) 756. Serne, R. J., in: Biological Implications of Metals in the Environment. Avail. as NTIS

CONF-750929, U.S. NTIS, Springfield, VA, 1977, p. 270. Schwertmann, W., Z. Pflanzenernaehr. Dueng. Bodenkd, 105 (1964) 194. Sholkovitz, E. R., Geochim. Cosmochim. Acta, 40 (1976) 831. Thomson, E., S. N. Luoma, D. J. Cain, and C. Johanson, Water, Air, Soil Pollut., in press.

the science of the total environment, 17 (1981) 165--196

Documents